Imposing codes of good conduct promotes social behaviour

Transcription

1 Imposing codes of good conduct promotes social behaviour Ann-Kathrin Koessler, Uwe Dulleck, Lionel Page October 31, 2014 Abstract In this paper we study experimentally how public statements about future social behaviour, used in oaths and code of conducts, affect individual decisions. Our results indicate that public statements of good intentions can help to overcome social dilemmas. We distinguish between voluntary and compulsory statements. Three effects can be identified: a selection effect that naturally occurs in the voluntary treatment for actors who choose to make the statement, a commitment effect due to the promise made and a coordination effect. The comparison of voluntary and compulsory statements shows, that the effect is strongest when the statement is imposed on everyone. keywords: public statement, oath, promise, commitment, public good, The authors would like to thank Uri Gneezy, Pedro Rey-Biel, Nicholas Jaquemet, Benno Torgler, Sarah Necker and Antoine Malezieux for very helpful comments on earlier versions of this paper. The authors also thank the participants at the French Experimental Economics Association conference, the Workshop on employees dishonesty at ESC Dijon for valuable comments. 1

2 1 Introduction Professional misconduct, for example by bankers or doctors, may seem at first like simple criminal behavior. The detection of respective breaches carries out legal consequences for the actor. But misconduct has a second, wider dimension. The detection of unethical behavior undermines the integrity of the entire profession and destroys public trust. Thus it is the common interest to preserve trustworthiness in the profession, but each individual profession member has an incentive to deviate from the professional standards for own benefits and free-ride on the compliance of the other profession members. The compliance to academic norms can be another example of such a trust common. An individual researcher can benefit from non-compliant behaviour, whether it is the case of serious misconduct as manipulating experimental data, or the weaker form of misconduct such as an incomplete description of experimental setups or only a partial revelation of research methods. But, in all cases, the private benefit comes with a potential cost to all researchers in the field. In case of detection the entire scientific community loses its credibility [for a general discussion on scientific fraud see Ioannidis, 2005; Martinson et al., 2005; Fanelli, 2009] [Begley and Ellis, 2012, for unreproducible landmark studies in cancer research][for a discussion in Economic research List et al., 2001; Necker, 2014]. In practice, a common approach to deal with this problem is to require oral and written statements promising proper, pro-social conduct from actors in the respective field. Statements of this kind are either voluntary or compulsory. Examples of compulsory statements are the Hippocratic oath of doctors, statements on conflicts of interest that are compulsory for many academic journals, or the MBA oath in some business schools. Recently, such oaths have also been proposed for bankers [Boatright, 2013] and economists [DeMartino, 2010]. Compared with the broad practical use surprisingly little research exists on the effect oaths and code of conducts[de Bruin and Dolfsma, 2013]. Our study aims to fill this gap and investigates experimentally how public statements about good intentions affect the behaviour in a social dilemma. The study is relates hereby to the existing literature in three ways. First, it is known from classic studies in Psychology that when individuals give a public commitment about a certain behaviour, they are more likely to perform this behaviour in the future [Festinger, 1957; Kiesler, 1971; Schlesinger, 2011]. Second, research in Behavioural Economics provides evidence that individuals are willing to keep a promise even if they have to forego a material gain [Gneezy, 2005]. It has been argued that this is either due to lying aversion [Ellingsen and Johannesson, 2004] or to a preference for promise 2

3 keeping per se [Vanberg, 2008]. The studies focus on one-to one interactions and how promises promote trust and trustworthiness. Third, it has been shown that the implementation of a solemn oath helps to avoid a hypothetical bias in experiments. Participants are more likely to reveal their true preferences [Jacquemet et al., 2013a]. And the solemn oath promotes trustworthy communication in experiments so that coordination failure can be overcome [Jacquemet et al., 2013b]. While the literature argues that voluntarily given promises have a stronger commitment effect [Kiesler, 1971], miss all existing studies to control for endogenity. An individual who is more inclined to behave socially or honestly may also more likely to choose to make a promise or oath. The observed increase in social behaviour may then be solely due to a selection effect. Our study identifies and controls for the endogenity with a within subject design. We answer the question whether, an additional increase in pro-social behaviour can be observed after the public pledge? A second difference to the existing literature is the audience of our promise. While existing promise studies focus on one-to-one interactions, we investigate in a collective setting whether public commitments can help to overcome social dilemmas. The promise is hereby directed to a group. We compare the effect of voluntary and compulsory statements of good intentions and test whether the effect of public commitment differs when the statement is voluntary or compulsory? After identifying and separating a selection effect, we find that public pledges do have a positive effect on prosocial behaviour. We call this commitment effect. Our results also show that the effect is strongest when the public pledge is compulsory. We see the reason in an assurance for the individual that the other actors face the same commitment. We call this coordination effect. The paper is organized as follows. Section 2 describes the experimental design and lists hypotheses about expected behaviour. Section 3 describes the results of the experiment. Section 4 concludes and derives policy implications. 3

4 2 Experimental Design In our experiment we use the classic public good game, in which participants are matched in groups of four and each player is endowed with 20 monetary units. Players have to decide how many units to contribute to the commons and how many units to keep for the private benefit. The social dilemma arises from the fact that the collective output is greater when all players contribute the entire endowment. But if players are selfish, they have an interest to keep all their units and free-ride on the others contributions to the common (see supplementary material for a precise description of the game). We allocated participants in three treatment groups. Table 1 summarizes the design. Every participant went through three different stages of the experiment. Each stage consisted of ten rounds of the public good game. After each stage the participants were assigned to a new group of players. The first stage was a baseline treatment. The control group repeated this baseline game in Stage 2, learning effects and heterogeneity between players could be so identified. For the voluntary oath group, Stage 2 introduced the possibility to publicly state the intention to contribute a minimum of 75% of their endowment. Players decided simultaneously whether they wanted to pledge this statement. Before the contribution decision all players learned how many other group member made the statement before they were asked to make contributions in Stage 2. In a compulsory oath group, the statement was compulsory to everyone. In the last stage, the same task was given but the players had now the option to punish other group members after they learned how much each players contributed to the common. In order to identify the pure effect of punishment, half of the groups in the control group did not have the option to punish other group members in Stage 3.All decisions were payment relevant, please see for details the supplementary materials Figure 1 provides an illustration of the game. 4

5 Table 1: Experimental design CONTROL VOLUNTARY COMPULSORY STAGE 1 Standard PGG Standard PGG Standard PGG STAGE 2 Standard PGG Voluntary oath Standard PGG Compulsory oath Standard PGG STAGE 3 Voluntary oath Compulsory oath Standard Standard PGG Standard PGG Standard PGG PGG Punishment Punishment Punishment Figure 1: Illustration of Game 5

6 Hypotheses If individuals are rational and selfish, as often assumed in standard economic models, they will contribute nothing to the common pool. Furthermore, any form of communication is meaningless and participants will not be affected by the statement. Ample evidence exists in the literature that participants are not behaving in line with this prediction. In experiments contributions to the common pool are on average positive, whereby people differ with respect to their contributions [Fischbacher et al., 2001]. Over the course of the experiment contributions deteriorate [Ledyard, 1995; Fehr and Gächter, 2000a; Chaudhuri, 2011]. It has also been found that communication enhances the contribution level [for an overview on the effect of communication Sally, 1995; Bochet et al., 2006; Balliet, 2009]. With respect to the statement, literature in Social Psychology shows that statements affect behaviour, even more when they are made voluntarily. Two theoretical explanations exist for this behaviour. One rests on the assumption that an aversion to lying exists, either because the person has a preference for keeping their word [Ellingsen and Johannesson, 2004; Vanberg, 2008] or because the promisor does not want to go against the social norm of not breaking a promise [Binmore, 2006; Bicchieri and Lev-On, 2007]. Alternatively, authors argue that the effect is more indirect: the statement raises the expectations of others, the person making the statement anticipates this and is motivated not to disappoint the expectations [Charness and Dufwenberg, 2006]. Both theories suggest that making a statement should increase contributions to the common in our experiment. Therefore, the first hypothesis to test is whether the statement makes a difference on the behaviour that can be observed: participants who choose to (voluntary treatment) or have to (compulsory treatment) make a statement contribute more than those that do not. In the case of voluntary statements, an effect could come from two potential explanations: First, a selection effect indicating that people who are intrinsically motivated to contribute more, tend to make the statement; Second, a commitment effect such that the statement itself increases contributions. Given our experimental design, the voluntary treatment group allows us to disentangle the commitment effect from the selection effect by comparing the Stage 1 behaviour to the Stage 2 behaviour. We can see whether participants who voluntarily choose to make the statement in Stage 2, have on average higher contribution in Stage 1 (selection effect). And we can determine if participants who voluntarily make the statement, increase their contributions in Stage 2 compared to Stage 1 (commitment effect). Subjects decisions might not only be influenced by the commitment, but also by the 6

7 information given in the offered statement. Hence (..)to contribute a minimum of 75% of the endowment may serve as reference point for desired behaviour. We therefore test whether the presentation of the statement itself has a framing impact on the situation. Respectively do participants who choose not to make the statement change their contribution behaviour in Stage 2 compared to Stage 1? Considering treatment group 2: The compulsory statement allows to study a possible coordination effect, which would subsequently give further insights into the commitment effect of a statement. In the literature it is shown that subjects adjust their contributions based on the contributions of other group members. This is referred to as conditional reciprocity, a well-recognised driver for contributions to the commons [Fischbacher et al., 2001]. If a statement affects the behaviour through a commitment effect, independent of whether the statement is voluntary or compulsory, then conditional reciprocity implies that contributions in the compulsory treatment group should be higher due to an additional coordination effect. Forcing all participants to make the statement triggers positive conditional reciprocity: subjects engage in more pro-social behaviour in situations when they expect others to do so as well. An open, empirical question is to what extent the commitment effect of a voluntary statement is stronger than of a compulsory statement. Observations of voluntary statement treatment, where all four group members chose to make the statement, may give some insights. One would expect that contributions in the respective group are higher than in the compulsory treatment groups. Choosing the statement voluntarily induces a stronger commitment [Kiesler, 1971; Charness and Dufwenberg, 2010] in these groups that complements a high coordination component. Stage 3 of the experiment introduces a punishment option to two baseline treatment groups, which may help to enforce cooperation. Even if groups were newly assorted, participants have a long history of play in Stage 3 given the two previous stages. This is likely to affect the play, and predictions as well as results regarding this stage should be treated with caution and seen only as indicative. Nonetheless, the ability to punish has been shown to increase cooperation in previous experiments [Fehr and Gächter, 2000a, 2002; Bochet et al., 2006], but can also trigger pre-emptive retaliation [Fehr and Gächter, 2000b; Herrmann et al., 2008; Nikiforakis, 2008]. Participants with low contributions retaliated against high contributors, trying to deter future punishment. But if people do not like to see their expectation disappointed, they might want to punish in our study. And punishment should be even stronger in the compulsory treatment group where the statement creates a common norm: we expect that participants who make low contributions and 7

8 made a statement, either voluntary or compulsory, are more likely to attract punishment (broken-promise punishment). And we predict that the enforcement of contributions in compulsory treatment group is more successful than in the other treatment groups. 3 Results We first discuss the effects of the statements without taking potential punishment into account, i.e. we consider Stage 1 and 2 of the experiment. Figure 2 shows averages and density of contributions in the different stages for the three treatment groups. We see that contributions for both treatments are significantly higher in Stage 2 than in Stage 1 (t-test: Voluntary Stage 1 vs. Stage 2: p < 0.01; Compulsory Stage 1 vs. Stage 2: p = 0.05). Additional the contributions in the compulsory treatment group is higher than in the voluntary treatment group, an indication for our coordination effect (t-test: Stage 2 Voluntary vs. Compulsory: p = 0.06). However the treatment comparison is impeded by differences already in the baseline between the three treatment groups. The standard deviation of average contributions is relatively large and some of this effect is due to variations across different experimental sessions. For this reason, we use the difference between contributions when looking at the effect of statements. Non-parametric tests are used for the ease of presentation, but the results are robust in regression analysis, which can be found in the supplementary material. If we distinguish between participants who voluntarily pledged the statement and others who did not, we find evidence for the selection effect. Figure 3 displays the average contributions for the respective groups. Average contributions of participants who chose to make a voluntary statement in Stage 2, are significantly higher (t-test, p < 0.001) in Stage 1 than of participants who decided not to make a statement. We also find evidence for the commitment effect. Participants who voluntarily or compulsory made the statement, significantly increase their contributions afterwards, relative to the baseline stage (t-test, p < 0.001). Subjects who decided not take the voluntary statement did not change their contribution level (t-test, p = 0.92). Figure 4 shows the dynamic development of contributions in Stage 2. To accommodate for differences between sessions we consider the differences between Stage 1 and 2. For the ease of presentation we use fitted values (see the supplementary material for the raw data). The effect of a voluntary statement is, in the beginning, as strong as of the compulsory statement, but the effect deteriorates faster when the statement was voluntary. The dynamic 8

9 Figure 2: Contributions in Stage 1 and 2 is driven by participants who made the statement and reduce their initial high contributions over time. Note however, that subjects who chose to make a voluntary statement were in all, but two groups, not in a situation where all four group members chose to make a statement. Contrarily for groups with a compulsory statement, it was always the case that all other group members made the statement. The faster deterioration of contributions in the voluntary treatment compared to the compulsory treatment, provides evidence for a coordination effect. Indicative of a difference in a commitment effect between the voluntary statement (treatment group 1) and a compulsory statement (treatment group 2) is the data of two experimental groups in the voluntary statement treatment, where all four group members chose to make a statement. In this case contributions stay higher for longer (a graph with the respective contributions can be found in the supplementary material). After the contribution choice participants were asked to indicate their expectations about the contributions of others (first order beliefs) and their guess of others expectations with respect to their own contributions (second order beliefs). The accuracy of guesses was incentivised (See SOM for de- 9

10 Figure 3: Contributions in Stage 1 and 2, statement tails). The results show that making a statement increased expectations of others, independent of whether the statement was voluntary or compulsory. The belief towards a person who chose to make the statement voluntarily was significantly higher compared to a subjects who decided not to make the statement (t-test, p < 0.001). Also in the compulsory treatment group, expectations are higher than in the control group or compared to baseline stage (t-test, p < 0.001). The rise in expectations is correctly anticipated by the person making the statement, the first order beliefs from the other players and own second order beliefs are highly correlated. Second order beliefs explain 70% of the variation in contributions made by a person that made 10

11 Figure 4: Fitted values for Difference between Stage 1 and 2 the statement. In the SOM we present regression results supporting these findings, allowing for more control in our analysis. Enforcement of Statements In a third stage of our experiment we allowed participants to distribute punishment points. Participants were able to see whether others in their group made a statement and how much they contributed in the current round. In the supplementary material we present a detailed analysis of the contributions and punishments choices of participants. The important findings are: that punishment points are usually awarded by high contributors to low contributors, and to a lesser extent from low contributors to high contributors. This is consistent with the literature [Fehr and Gächter, 2000a,b]. Comparing the treatments, we observe that subjects in the compulsory treatment receive harsher punishment for contributions below the stated level of 15 ECUs. In the voluntary treatment group, there is no significant disparity between participants who made a statement to participants who decided not to make a statement (see SOM). When considering the total social output, punishment was effective in the voluntary statement group. It increases contributions, but from a social perspective a lower welfare level was realised than in the Stage 2 without enforcement. The punishment points reduce the overall surplus more than 11

12 they increase contributions. Only for the compulsory treatment group the welfare level was higher in Stage 3 with punishment than in Stage 2 without punishment. The introduction of punishment lead to higher contributions and the social cost of punishment was less than the gain from increased contributions. 4 Conclusions Our results indicate that public statements of good intention, used in oaths and codes of good conduct, can help to overcome social dilemmas. Rational choice predicts that such statements are merely cheap talk and the effect is, if at all due to a selection effect. We show that this is not the case: our results provide evidence for an additional commitment effect. Through a within subject design we controlled for endogenity. Once a statement has been made it has a positive effect on the level of contribution. We have also found evidence for a coordination effect with higher contributions by an individual if all group members make a statement, independent whether the statement is compulsory or voluntary. This observation suggests that compulsory, and not voluntary, statements have a stronger effect on the overall level of contributions to the commons. When the public statement was compulsory, subjects also kept their contributions higher for longer. To what extent are the results specific to the artificial situation in an experimental laboratory and what can be taken as implications for the real world? It can be critizised that the contributions to commons are highly artificial and misconduct is clearly defined. In natural environments the lines are not as clear. However, the purity creates controllability in our experiment. Thus we can control for endogenity and can identify the pure effect of the statements. The moral or emotional loading of the statements are hereby substantially lower than in a real world setting. We expect the effect to be stronger in the natural environment. We are also aware that an experimenter effect may exist. But we expect that this effect is similar for the baseline as well as the treatment groups. Thus if the statements do not affect the experimenter effect, our analysis is able to shed some lights on the effect of public statements on contributions to the commons. Our results suggest that public statements of good intentions can help to promote professional integrity. Our findings support policies that request statements about conflict of interest or ethical conduct, as required for example to enter a profession or for submission of an article to an academic journal. In general, our research suggests that even a non-binding statement of intention to contribute to the commons, increases the overall level of contributions. Thus, where it may be 12

13 politically difficult to get agreement to legally binding rules and regulations that ensure contributions to the commons, public statements may be a less invasive and politically easier solution to mitigate the problem. 13

14 References Balliet, D. (2009). Communication and cooperation in social dilemmas: A meta-analytic review. Journal of Conflict Resolution, 54(1): Begley, C. G. and Ellis, L. M. (2012). Drug development: Raise standards for preclinical cancer research. Nature, 483(7391): Bicchieri, C. and Lev-On, A. (2007). Computer-mediated communication and cooperation in social dilemmas: an experimental analysis. Politics, Philosophy & Economics, 6(2): Binmore, K. (2006). Why do people cooperate? Politics, Philosophy & Economics, 5(1): Boatright, J. R. (2013). Swearing to be virtuous: The prospects of a banker s oath. Review of Social Economy, 71(2): Bochet, O., Page, T., and Putterman, L. (2006). Communication and punishment in voluntary contribution experiments. Journal of Economic Behavior & Organization, 60(1): Charness, G. and Dufwenberg, M. (2006). Promises and partnership. Econometrica, 74(6): Charness, G. and Dufwenberg, M. (2010). Bare promises: An experiment. Economics Letters, 107(2): Chaudhuri, A. (2011). Sustaining cooperation in laboratory public goods experiments: a selective survey of the literature. Experimental Economics, 14(1): de Bruin, B. and Dolfsma, W. (2013). Oaths and codes in economics and businessintroducing the special issue. Review of Social Economy, 71(2): DeMartino, G. F. (2010). The Economist s Oath: On the need for and content of professional economic ethics. Oxford University Press. Ellingsen, T. and Johannesson, M. (2004). Promises, threats and fairness. The Economic Journal, 114(495): Fanelli, D. (2009). How many scientists fabricate and falsify research? a systematic review and meta-analysis of survey data. PloS one, 4(5):e

15 Fehr, E. and Gächter, S. (2000a). Cooperation and punishment in public goods experiments. American Economic Review, 90(4): Fehr, E. and Gächter, S. (2000b). Fairness and retaliation: The economics of reciprocity. The Journal of Economic Perspectives, 14(3): Fehr, E. and Gächter, S. (2002). Altruistic punishment in humans. Nature, 415(6868): Festinger, L. (1957). A Theory of Cognitive Dissonance, volume 1. Stanford university press. Fischbacher, U., Gächter, S., and Fehr, E. (2001). Are people conditionally cooperative? evidence from a public goods experiment. Economics Letters, 71(3): Gneezy, U. (2005). Deception: The role of consequences. The American Economic Review, 95(1): Herrmann, B., Thöni, C., and Gächter, S. (2008). Antisocial punishment across societies. Science, 319(5868): Ioannidis, J. P. (2005). Why most published research findings are false. PLoS medicine, 2(8):e124. Jacquemet, N., Joule, R.-V., Luchini, S., and Shogren, J. F. (2013a). Preference elicitation under oath. Journal of Environmental Economics and Management, 65(1): Jacquemet, N., Luchini, S., Shogren, J. F., and Zylbersztejn, A. (2013b). Coordination with communication under oath. Kiesler, C. A. (1971). The psychology of commitment: Experiments linking behavior to belief. Academic Press New York. Ledyard, J. O. (1995). Public Goods: A Survey of Experimental Research. Princeton: Princeton University Press. List, J. A., Bailey, C. D., Euzent, P. J., and Martin, T. L. (2001). Academic economists behaving badly? a survey on three areas of unethical behavior. Economic Inquiry, 39(1): Martinson, B. C., Anderson, M. S., and De Vries, R. (2005). behaving badly. Nature, 435(7043): Scientists 15

16 Necker, S. (2014). Scientific misbehavior in economics. Research Policy. Nikiforakis, N. (2008). Punishment and counter-punishment in public good games: Can we really govern ourselves? Journal of Public Economics, 92(1): Sally, D. (1995). Conversation and cooperation in social dilemmas a metaanalysis of experiments from 1958 to Rationality and society, 7(1): Schlesinger, H. J. (2011). Promises, oaths, and vows: on the psychology of promising. Taylor & Francis. Vanberg, C. (2008). Why do people keep their promises? an experimental test of two explanations1. Econometrica, 76(6):